Author: Miguel Albaladejo, Ana Corbalan, Aitor Ortega, Vicente Puchades, David Ramos, Alfredo Serna-Berna, Jonattan Suarez 👨🔬
Affiliation: Hospital General Universitario Santa Lucia 🌍
Purpose:
Prostate SBRT treatments are frequently delivered using standard VMAT IGRT technique. The aim of this study is to test the feasibility of Ethos Artificial Intelligence (AI) prostate SBRT by comparing AI-generated (V-2.1) plans with Eclipse (AXB V-16.1.0) standard plans at our department.
Methods:
Four prostate SBRT patients (40 Gy in 5 fractions to PTV) were used for this analysis. AI-plans consisted of 9 IMRT sliding window fields 6xFFF, while Eclipse plans were 2-3 full-arc VMAT fieldsv6xFFF. As the Ethos planning system uses a 2.5mm dose grid, AI-plans were recalculated in Eclipse (1.2mm grid resolution). Additionally, a hydrogel spacer was used for rectum sparing in all patients.
Clinical dose-volume metrics (DVM) were analysed for GTV (Dmean, V39.20Gy), PTV=GTV+3mm posterior+5mm all other directions (V39.20Gy, V42Gy), Rectum (D1cc, Dmax, Dmean, V10Gy, V24Gy, V30.15Gy), Bladder (Dmax, D10cc, D50%), Urethra-PRV=Urethra+2mm (D1cc, Dmax) were evaluated for all plans.
Results:
Although no statistically significant differences in DVM were found between Eclipse and AI-plans, several trends were observed in the median DVM:
All rectum DVM were lower in Eclipse plans compared to AI-plans, particularly for mean dose (9.4 Gy Eclipse vs. 10.6 AI).
Bladder D10cc (30.4 Gy Eclipse vs. 34.7 Gy AI) and D50% (1.4 Gy Eclipse vs. 1.7 Gy AI) were lower in Eclipse plans.
Urethra PRV maximum dose was notably lower in AI-plans (40.6 Gy AI vs. 41.3 Gy Eclipse). However, AI-plans achieved higher PTV coverage at V39.20Gy (99.68% AI vs. 98.84% Eclipse).
Conclusion:
Both Ethos AI and Eclipse plans met our clinical dosimetric goals, and furthermore, AI plans achieved better PTV V39.20Gy while reducing maximum dose to Urethra PRV. Thus, we consider that AI-planning can be used as online adaptive prostate SBRT.